Semantic variability predicts neural variability of object concepts.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Linguistics

Background:

• Traditional neuroscientific concept study averages neural patterns, treating variation as noise.
• This contrasts with other methods that acknowledge variability in conceptual processing.
• The current study questions whether neural pattern variation reflects true contextual variability.

Purpose of the Study:

• To investigate if neural pattern variation during semantic retrieval exceeds measurement error.
• To determine if this variation reflects contextual variability.
• To link semantic variability (SV) to neural variability.

Main Methods:

• Quantified semantic variability (SV) using word frequency and co-occurrence in large text corpora.
• Conducted functional magnetic resonance imaging (fMRI) to measure neural activity.
• Sampled neural activity for each concept across three distinct, randomized contexts.

Main Results:

• A concept's semantic variability (SV) score significantly predicted its neural variability.
• Concepts with higher SV showed more dynamic neural representations.
• Concepts with lower SV exhibited more uniform activation patterns.

Conclusions:

• Neural patterns evoked by semantic retrieval are not static but context-dependent.
• Findings support a flexible, distributed model of semantic memory organization.
• Both concept meaning and neural activity patterns dynamically vary across contexts.